[BUG] Plan execution interrupted when editing queued message during Plan mode

[Rokurolize]

Environment

• Platform (select one):
  • [x] Anthropic API
  • [ ] AWS Bedrock
  • [ ] Google Vertex AI
  • [ ] Other:
• Claude CLI version: 1.0.25 (Claude Code)
• Operating System: Windows 11 (WSL)
• Terminal:

Bug Description

When using Plan mode, if a user sends a message immediately after selecting "1. Yes" to execute a plan, and then edits that queued message using the Up key, Claude abandons the current execution and re-presents the plan instead of continuing with the original task.

Steps to Reproduce

1. Enter Plan mode and receive a plan with options "1. Yes" or "2. keep planning"
2. Select "1. Yes" by pressing Enter to start plan execution
3. Immediately send a message after selecting option 1
4. Wait for the message to appear in the queue on screen
5. Press the Up key to edit the queued message
6. Confirm the edited message

Expected Behavior

Claude should continue executing the original plan while handling the additional message as supplementary input, or at minimum, ask for confirmation before abandoning the current work.

Actual Behavior

Claude will completely halt the current plan execution and re-present all planning options from scratch. This means Claude will start fresh with reading the current code, effectively discarding any work it had previously implemented for the original task.

Additional Context

This behavior suggests that selecting "2. keep planning" might be the intended workflow when users want to add additional context or instructions to a plan. The current behavior could lead to lost work and confusion about the proper workflow.